Software Engineering and Information Technology

The following sections are included:

• Editorial
• Contents

Nodes in a sensor networks may be lost because of power exhaustion or malicious attacks. Therefore it's necessary for new nodes' deployment. Using Bloom Filter, we propose an access control protocol including the mutual authentication procedure and common key generation. It's efficient for the nodes' addition, deletion and authentication. At last, the paper makes an analysis of the performance and safety about the proposed protocol.

Due to the exponential growth of mobile data traffic, Heterogeneous Cellular Network (HCN) has been a long-term trend of mobile communications, which will also bring rapid increase of energy consumption. In this paper, the average achievable capacity of users in a cell and the throughput of users at the Macro cell-edge are derived by using a stochastic geometry based model. With consideration of mobile traffic model, a technical approach of Pico base station (PBS) sleep mode is introduced in order to achieve the lowest energy consumption. An energy consumption minimization problem is formulated and the optimal sleeping probability for PBS is determined. Numerical results confirm the achieved energy savings performance of proposed sleep mode without damaging the system performance.

The explosive increasing books puzzle the librarians in managing, sorting, circulating and inventing them efficiently and accurately. The widely used bar codes technology in the libraries is unable to meet the requirements. Considering the advantages of RFID technology such as contactless penetrating tag identification, large information capacity of tag, The paper developed a library management system using the RFID technology under the Qt Creator integrated development environment, which uses RFID tags and readers to realize the library information inputting, lending and return and library information management, etc.

According to the different requirements of clock synchronization's precision for intelligent substation system at all layers, this paper has applied the GPS clock synchronization technology and PTP clock synchronization technology to intelligent substation system's simulation platform. It has designed a clock synchronization scheme which can reach to microsecond precision by using substation control layer, bay layer and process layers three layers hierarchical clock synchronization mechanism, so as to ensure the equipment's clock being accurate, reliable, and unified at all layers of the intelligent substation system.

In this paper, we are interested in designing a transmission scheduling algorithms in a packet radio network which can satisfy the requirements of such ad-hoc, mobile multimedia applications. Firstly the impact factors of packet collision are analyzed from two viewpoints of broadcast and point to point. Secondly a complete characterization of the link and broadcast scheduling problems is given, including a full analysis of the interference environment proper to each. Furthermore, a comprehensive comparison of link and broadcast scheduling is presented, including a simulation to demonstrate the throughput characteristics of each algorithm.

Single feature unvoiced / voiced classification algorithm usually not only fails to provide more accurate unvoiced / voiced information, but also the robustness to noise is low, which easily results in the speech quality of vocoder decoding become poor. To overcome this problem, one kind of linear discriminant algorithm of minimum square error criterion based on multi-feature joint is put forward. The classification of unvoiced / voiced is seen as two categories of mode recognition problems in the algorithm, sample vector of short-time speech is made up of three features of short-time amplitude, short-time zero-crossing rate and short-time self-correlation maximum value of speech; normalized augmented sample matrix is constructed by normalized augmented sample vectors of current frame and former four frames, and classification labels of unvoiced / voiced of sample vectors are added automatically; special margin vectors are constructed by constant that is matched with sample matrix; average vector of overall sample is calculated by sample matrix and classification labels; gradient descent algorithm of minimum square error criterion function is adopted, and optimal augmented weight vector of short-time speech is obtained; therefore, the classification of unvoiced / voiced is judged. The minimum memory and automatic update with the frame of all parameters are realized in the study, thus achieving inter-frame self-adaption of study and judgment. Simulation experiment shows: both the accuracy of classification of unvoiced / voiced and the robustness to noise have prominent enhancement than classical single feature method.

This paper constructs frame design and work process of indoor positioning system for iBeacon Bluetooth, it puts forward system method with high-precision indoor positioning of self-adaptation learning and algorithm. It makes analysis and points out that DSK-SVR algorithm positioning method design can extract the judgment ability of positioning characteristic from the perspective of enhancing generalization of learning machine.

Web storage is an important technique to scale the performance optimizing of front-end website. Due to the limited space in web storage, exceptions would occur and nothing can be stored when the space is full. A low overhead and adaptive method based on Least Recently Used-Threshold (LRU-Threshold) algorithm is proposed in this paper to solve the problems. It generally considers recency as the main factor and chooses the threshold on the basic of the reading rate of items. Some items would be replaced with a new one according to the location of reference seen in the web storage streams and some will not be stored over the threshold. Through the experiments, the data in web storage can be replaced automatically and exceptions would not occur.

Considering the deficiency of traditional HF ALE system on frequency usage, integrating the concept of cognitive radio technology, this paper puts forward a channel availability prediction method for HF communications based on Hidden Markov Model (HMM). We set up a three-state HMM prediction model of channel availability according to the principle of channel usage in HF communication, and give out the corresponding specific prediction process. The simulation results show that the model can well predict the changes of channel with accuracy above 89%.

The “non-consensus” opinions pervade all patterns of human's interactive activities. In this paper, we study the phenomenon of the information competition dynamics induced by messages with opposite meaning. We construct a new information competition model which could perfectly match the reality in complex networks. It is shown that the k-shell value of nodes plays an important role in reflecting its competitive power in the information competition processes, classifying the nodes in the network with the k-shell decomposition technique. In particular, by varying the variables in our model different cases of information competition phenomena can be successfully explained. Our findings indicate that this new information competition model should be useful to the study on network information dissemination.

On-demand service is a new technology in cloud computing. The resource demand, which is changing dynamically, is the input of the optimal reliability design. So predicting the resource demand becomes very important in cloud computing. In this paper, we propose a resource demand predicting method based on cluster analysis and neural network, which divides the demand into several groups, and provides detailed prediction for different data group. This method is verified through data collected from ClarkNet. The results show that our proposed method is more effective and accuracy.

We propose a remote control using radio waves of new traffic lights traffic command mode, to solve the irrational problem of traffic congestion crossing when the vehicle from different directions. In the paper, we propose the traffic lights remote control system based on the use of DF data transceiver module, using PT2262 / PT2272 as codec chip for long distance control of traffic signals. Traffic police could change the traffic lights countdown time timely using their remote controls based on the changes in traffic flows at different times. This approach can replace the sign directing traffic, but more simple and convenient. It uses a micro control unit MCU, and the DF data transceiver module to transmit and receive instruction. This is a kind of radio wave communication in the 100m range farther than the infrared remote control, which could satisfy the actual needs. This system enables automatic operation and temporary intervention of traffic lights combined with stable and reliable. This system can drive lights and timing display circuit directly, it has a strong practical value.

The satellite network is composed of several satellite and ground stations pointing to the satellite. The satellite constellation network plays an increasingly important role in the development of space. The inter satellite link technology which can guarantee the satellite network is a new trend of development. In this paper, a method of mutual visibility analysis of constellation internal satellites is established, and the validity of the method is demonstrated by simulation.

A kind of safety quality management system of the engineering construction enterprise is designed, including site Smartphone terminal and the background safety quality management information system. The Smartphone terminal is applied with ARM9 core MCU to control as well as IPv6 wireless self-organized network and GPRS technology for communication. The Smartphone terminal sends collected safety quality text and media information in the construction site to the aggregation node through wireless network and then the aggregation node sends data to the background safety quality management information system through GPRS network. The safety management personnel of the bureau forms the statistical data according to the declared safety and quality management data of the engineering project as well as issue it to the corresponding site terminal equipment according to the early warning level.

Function P-sets (packet sets) is a novel model with law and dynamic characteristics, which is composed of function internal P-set and function outer P-set S^F. Using the structure, law and dynamic characteristic of function P-sets, the band information law and their generation as well as the hiding and hiding theorem of P-information in band information law and the recovery theorem of information law are given. Using the above theory results, an application example of information law hiding is given.

Agricultural scientific technical information service is an important bridge between agricultural technology achievement information and agricultural production. By the investigation of the current situation of agricultural scientific technical information consulting services in Beijing, and the analysis of restricting factors of the service, the technical countermeasures of its development under the new situation are provided in this paper to give the reverence for the construction of new agricultural scientific technical information counseling service system.

The security of the partially blind signature scheme without certificates proposed by He Junjie et al is analyzed, it is found that the scheme can not resist tampering attack on negotiated information. Therefore, proposed a partially blind signature scheme which can resist tampering attack on negotiated information, and the new scheme is proved to be existentially unforgeable against adaptive chosen message in the random oracle model and inv-CDH assumption difficulties.

The partially blind signature is an extension of the blind signature, which allows signer to include some common information negotiated by signer and the user of requesting signature. An efficient certificateless partially blind signature scheme is constructed. It is based on identity-based signature scheme proposed by Shim. The analysis of its correctness, partial blindness are subsequently presented. The new scheme based on the Computational Diffie-Hellman Problem (CDHP) and CDHP assumption. The new scheme based on the CDHP assumption is proved to be existentially unforgeable for adaptive chosen message and identity attacks in random oracle model. To analyze the performance of the scheme, indicating that the new scheme have more advantage than the majority of the existing partially blind signature scheme.

Authentication in multi-server environment plays an important role in distributed systems. Multi-server authenticated key agreement scheme allows users to register only once, and then access multiple services with session key establishment. However, security and privacy are crucial issues to guarantee identity authentication, data confidentiality and participant privacy. In this paper, we cryptanalyze Mishra et al.'s ‘A secure user anonymity-preserving biometric-based multi-server authenticated key agreement scheme using smart cards’, and point out that their scheme is vulnerable to session key compromised attack by insider curious service servers, and user behavior traced attack by common malicious attackers. In order to remedy the above weaknesses, we propose an improved scheme to enhance the security guarantee and privacy protection. Our improvement introduces public key for service servers to safeguard user identity and session key freshness. It not only remains the merits of the original scheme, but also achieves more security and privacy requirements. Finally the analysis shows that our improved scheme is more practical, but there is a tradeoff between security/privacy and performance.

Because Web applications hold the feature of cross-platform and others, so more and more applications use this form, so that its development environment is gradually maturing and variety of stable and rapid development framework have also appeared, such as SSH framework, but Web application performance issue has been always a problem occurs during operation. This paper took the teaching Web application as the background, described and researched for the potential problems, proposed solutions, and finally tested it to the desire to provide reference for similar studies.

In this paper, using the modern information technology, on the one hand, establish crop breeding information intelligent collaborative system, and provides a technical support and general tool for all kinds of data obtained from the process of crop breeding information integration and analysis; on the other hand, as a kind of key technology research and exploration, using the multi-agent technology in the field of distributed artificial intelligence, puts forward a kind of intelligent, collaborative and sharing method about big data of crop breeding control and management, for the big data of crop breeding research provides a new research method and train of thought.

Nowadays a new big data era has begun. Blogs, Microblog, the Social Network and Internet of Things have made a wide variety of big data, there are great value in big data, but privacy protection is one of the major challenges of big data. In this paper, based on the personalized service demand of privacy protection data publishing, the latest progress of personalized privacy anonymity technology were reviewed. We analyzed and summarized the anonymous models according to the different forms of personalized needs such as CBK(L,K)-Anonymity and (k,l,p)-anonymity, and then we summarized the full paper and further research direction on personalized privacy anonymous technology was prospected.

With the rapid development of social networking and mobile e-commerce, a new e-commerce model of O2O mode quickly grows up. This mode is based on interactive of online and offline. And it is focuses on the experience of the model. Furthermore, this model expands a new way for marketing. This study aims to do the deep research on O2O mode, seeks to carry out innovative way and effect of O2O marketing model based on social networks.

In order to improve long term durability of bioprosthetic heart valve, we analyze and compare stress distribution of bioprosthetic heart valve leaflets with different thickness under the same load. We will establish the elliptical leaflets models via computer aided design. Based on the parametric models of the heart valve, three kinds of thickness (0.3 mm, 0.4 mm, and 0.5 mm) of BHV is analyzed by us and finite element analysis is used to simulate the mechanical performance of HBV during the diastolic phase. We can conclude that the bioprosthetic heart valve leaflet with different thickness has a significant effect on the dynamic behavior of the bioprosthetic heart valve. According to the whole loading process, we can conclude that the thickness of the elliptical valve leaflet with 0.4 mm is better than others. The finite element analysis on the HBV could provide reliable and useful testimony for the HBV manufacturing and optimizing.

The paper aims to analyze the problems of implementing ERP management system, and the implementation progress of the maintenance of some contradictory information, system security and other aspects. Equipped with the management of ERP system in production and business operation activities, it can help enterprises to improve the management performance and the implementation of ERP. This paper provides the corresponding technical solutions for enterprises. In reality, through the use of Virtual Desktop technology for centralized management and the use of the ERP system, enterprises can improve management performance.

Access control plays an important role in cloud storage. Attribute-based encryption enables the data owner embedded the access policy in the data. Many solutions have been proposed to ensure data security on cloud. However, in the existing indirectly revocation schemes, the efficiency and computational costs need to be improved. In order to realize fine-grained access control and reduce the computing burden of the data owner and users, we propose a scheme which includes attribute-based access control with revocation. The scheme could ensure data security, achieve higher computing efficiency and realize expressive policy.

The plant crown can be considered as fractal, how to describe the dense degree of plant crown by fractional dimension, there is a little research. In this paper, the Crown dimension of plant with trigeminal tree structure is calculated by box dimension method, the box dimension is .

Information disclosure system is the law that listed companies must obey in order to insure the benefits of investors and hold up to public scrutiny. Information like financial statement changes and operation condition of listed companies must be to the public according to the system. XBRL is an XML-based extensible language for exchanging business information. The language is widely used in financial information disclosure system and becomes the standard data format of the system. The specification, taxonomy and instance documents of XBRL are researched in this paper and the method of parallel data mining for the frequent pattern of massive XBRL data is proposed based on MapReduce and HDFS. The XBRL instances of the listed companies in China are processed by using this method and proved it to be effective.

For large-scale data processing, complex scale, and the large amount of data, this paper proposes a new parallel scheduling optimization framework and corresponding algorithms, to improve the actual application effect of data processing. Parallel scheduling framework which is suitable for the large-scale data processing designed in this paper, this framework can further improve the parallel data processing. The main feature of the framework is the task of classification, dynamic scheduling and task feedback information, to provide further support for dynamic adjustment of the whole framework. Through the cooperation of each module in the framework, we can realize the parallel processing of data and ensure the accuracy, the stability and efficiency of data processing.

An Ontology is a shared formal specification of a conceptual model. This paper stated the necessity of constructing travel ontology from the perspective of ontology, inspected the design of class structure in the case of Yang Zhou Slender West Lake. By OWL DL, one ontology description language, this paper defined the ontology classes, relationship and constraints between attributes of ontology classes, and defined creation of class instances to complete the construction of domain ontology knowledge base. The visualization of ontology was performed through the ontology editing tools protégé 4.2 and graphviz 2.28.

As the fuzzy C- means (FCM) clustering algorithm is easy to fall into local optimal value, this paper presents a new FCM clustering algorithm based on improved shuffled frog leaping algorithm (SFLA). The algorithm uses shuffled frog leaping algorithm for its fast global searching ability to generate initial clustering centers of FCM, and then use FCM to optimize the initial cluster centers. Finally we get the global optimal solution. In order to improve shuffled frog leaping algorithm ability to solve complex problems, Cauchy variation and chaos optimization mechanism is introduced in the frog update strategy, and improves the performance of shuffled frog leaping algorithm. In order to enhance the classification accuracy, a characteristic index weighting factor is introduced in the FCM, and the clustering accuracy and speed of FCM is improved. The results of experiment on real data show that searching ability and precision of the new algorithm (ISFLAFCM) are improved obviously contrast with the traditional FCM clustering algorithms.

Based on the data mining method, Support Vector Machine technology, Correlated Analytical Variable Selection method and Principal Component Analytical Variable Dimension Reduction method, setting five-category classification and prediction model for “Three rural” loan risk of commercial bank, then make classification prediction for an Agricultural Bank of China in Changsha and verify the outcome through this prediction model, and make comparative analysis with the classification prediction accuracy of neural network and the time consumed for processing, the advantage of this method is obvious. This five-category classification and prediction model for loan risk have great meanings for bank to establish more robust customer relationship management system and attract more high-quality customer resources.

In this paper, oscillation of even order half-liner neutral differential equation with distributed deviating arguments is studied. By means of yang inequality, the generalized Riccati transformation and the averaging technique, several sufficient conditions are obtained for oscillation of all solutions of the equation.

To explore consumption law of armored equipment maintenance material, and to guide reserve and serve issues, this paper studied the association rule mining of the equipment maintenance material consumption data. Two typical association rule mining algorithms are expatiated and compared based on Hadoop MapReduce. Taking a typical equipment maintenance material consumption data as an example, this paper illuminated the executive processes of algorithms and analyzed the mining results. The study can provide reliance to the consumption standards perfection of current equipment.

Polybrominated diphenyl ethers (PBDEs), used as an important class of additive flame retardants in polymers, post potential hazards to the environment, wildlife and humans, and it is thus essential to find an efficient way to eliminate the PBDEs contamination. The reductive debronimation pathway of BDE-47 and BDE-99 was investigated in this study. The stepwise debromination was shown as the dominant reaction process for PBDEs. The quantitative structure property relationship (QSPR) was also performed in this study to develop models for predicting the reactions of reductive debromination. And the predictive results from the QSPR studies show that the nonlinear model (back propagation artificial neural network (BPANN)) is better than the two linear models (principal component analysis-multiple linear regression (PCAMLR) and partial least square regression (PLSR)). For the BP-ANN model, we also considered the overfitting problem with limited number of hidden neurons.

Big Data has enabled businesses to monitor large amounts of data and extract pertinent patterns necessary for decision making. Food safety on the other hand, has necessitated the deployment of quality control (QC) systems like the hazard analysis and critical control points (HACCP) framework to reduce the amount of harmful pathogens that consumers are exposed to through food. However, the complexity of adapting Big Data to the food safety concern has exposed the lack of proper control of variables that dictate whether practitioners are in agreement on which aspects of food processing and supply chain management are most important to the control process. This research will propose a generic framework, which adapts Big Data to food safety risk monitoring systems more efficiently.

In recent years, the problem of the employment option of college graduates is paid more and more attention in the society. Every year the situation for the employment option of college graduates is faced with many uncertain factors. Therefore, the study on the employment of college graduates has very important practical significance. At present, we have accumulated a wealth of data about the employment options of college graduates in previous years. Data mining in handling high-volume data, excavate potential relationship has incomparable advantage. Therefore this paper proposes using bayesian data mining classification algorithm, through training the existing data of the employment option of college graduates, analyze the birth to the students graduated that were Satisfied with their jobs. And then we gain the classification feature set rules, and establish classification model about the employment option of college graduates. Using the model, we can divide the employment option of college graduates into three categories: the first that they were satisfied with their jobs, the second that they were General satisfied with their jobs, and the third that they were not satisfied with their jobs at all. Using the method of fuzzy mathematics and according the actual situations of the college graduates, we can calculate the inclusion degrees so as to determine which category they belong to. It is proved by the experiments that this model is selected user groups with high accuracy. The data of employment option of college graduates is of certain guidance.

Nowadays, the automatically learning Chinese word segmentation algorithms based on particle swarm optimization (PSO) were presented and discussed. However, most of the papers employed PSO as the parameter optimizer for neural networks. We proposed a novel pure PSO-based CWS schema (PSO-CWS) and tested the practicability on the bakeoff 2005 dataset. Firstly, the content is split into a mass of clauses without any punctuation. Then clause-centered PSO-CWS generates the PSO swarm in the segment searching space. Led by the self-knowledge and the social knowledge, PSO-CWS covers the best segmentation schema with the highest fitness. Experiment results show that PSO-CWS proposed in this paper achieved the acceptable accuracy of Chinese word segmentation and the practical efficiency.

This paper discusses the five kinds of swarm intelligence algorithm. On this basis, the interactive user interface simulation of swarm intelligent algorithm optimization function was designed with the function of GUI in MATLAB. This platform has the advantages of friendly interface, simple operation, strong interactivity, visualization, scalability etc. And any continuous function can be simulated, analysed and compared using the algorithm in the platform. At the same time, this paper proposed real coded chaotic quantum genetic algorithm based on catastrophe. The simulation results in the platform and improved algorithms are compared, which can get rid of the understanding of multiple swarm intelligent algorithm in the platform and save time. The simulation results are distinctive and highlight, and it shows that the improved algorithm has high precision, overcoming the premature and the advantages of fast convergence.

Ant colony optimization algorithm is applied to loading balancing mechanism of marine surveillance intelligent detection system to increase its monitoring efficiency. Pheromone updating role and routing role is optimized to increase concurrency performance, and load balancing architecture is designed for the system to implement the algorithm. Practical test results demonstrate the efficiency and performance of the ant colony optimization algorithm.

People need a fast and efficient modeling method to describe the 3D of Marine environment. Three-dimensional modeling provides a precondition for the path planning. This paper introduces a way for modeling three-dimensional with ant colony optimization and cuckoo search which has strong robustness. The selection of the initial line is not high. Then the hybrid optimization does not need manual adjustment in the process, and its parameter is relatively small. Some simulations and experiments in this paper show that the ways with ant colony optimization and cuckoo search have a good performance in modeling three-dimensional path planning, the path is shorter than other algorithm. It prove that the method is effective both in establishing the corresponding data space, combining with the data for the path planning.

This paper studies the problem of low delay oriented optimal sequential sensing order in cognitive small cell (CSC). CSC is a promising technology in the evolving 5G networks, which could opportunistically use the free channels unoccupied. The finding of idle channels could be carried out via spectrum sensing. We propose a sensing delay oriented fast sequential sensing order discovery algorithm. The capacities of channels, idle probabilities of channels, sensing delay and the target resource are jointly considered. The theoretic analysis gives out the sensing order decision condition. The computing complex of the proposed algorithm is also analyzed. Simulation results show that the proposed approach outperforms existing methods.

Surface roughness of an engineering surface is composed of short-wavelength components, and it is formed by subtracting long-wavelength components from the raw profile of an engineering surface. A mean line is commonly used to express long-wavelength components during surface evaluation. In our previous research, we proposed an EMD mean line to separate surface roughness profile from raw profile, but it suffers from some deviations at two ends of an evaluation length. This article tries to overcome the problem by using an EMD mixed mean line. Firstly, Empirical Mode Decomposition (EMD) is utilized to decompose the raw surface profile into a series of Intrinsic Mode Functions (IMFs) and a residue. Secondly, IMFs with frequency less than 10 within an evaluation length and the residue, are selected for reconstructing an EMD reference line. An Amplitude Weighting Factor (AWF) is further applied to the selected IMF with maximum frequency to form an EMD mean line. Finally, an EMD mixed mean line is created by mixing the middle segment of EMD mean line with the two boundary segments of EMD reference line. Varied Amplitude Weighting Factor (VAWF) is designed to complete and smooth the EMD mixed mean line. Experimental results show that the deviations at two ends are suppressed and the EMD mixed mean line is more accurate for roughness profile extraction.

Swarm animation is an important research direction in the field of computer animation, which aims at computer simulation of swarm behavior, real-time rendering of swarm animation and other relevant aspects. In these aspects, swarm path planning is a key issue and has received widespread attention. Some global path planning methods, such as A* and Dijkstra, due to need predefined environment information and high computation burden, is not suitable for swarm path planning; But some optimization algorithms based on swarm intelligence theory, such as particle swarm optimization (PSO), ant colony optimization (ACO) and artificial bee colony (ABC), etc. Because of its simplicity and fast convergence, has become the best choice of swarm path planning. This paper presents a multi-layer swarm path planning method. Firstly we divide the virtual environment into three layers, including geometry layer, topology layer and navigation layer. On the basis of this environmental model, the swarm path planning model is divided into inner layer and outer layer model, outer layer adopts an improved A* algorithm for topological path planning, inner layer adopts a hybrid algorithm based on ABC and PSO for dynamic path planning, then path data derived from two layers are joined together to form a final path result. Experimental results show this multi-layer swarm path planning method effectively reflected intelligence and authenticity of individuals, improved execution efficiency of the algorithm, and prevented premature convergence.

The neutral point clamped (NPC) pulse width modulation (PWM) inverter has been put into practical use for large capacity AC motor drives because of its less distorted output, lower costs and better control performance. This paper suggests a way to improve the performance of permanent magnet synchronous motor (PMSM) by using the three-level NPC inverter. Firstly, the three-level NPC inverter space voltage vector pulse width modulation (SVPWM) technique is analyzed in detail. Then, the NPC inverter is applied in PMSM vector control. Finally, a simulation experiment of the proposed control algorithm is carried out. The simulation results show that the proposed method can effectively suppress the torque ripple and improve driving performance for the PMSM drive.

Travelers' route choice behavior is usually based on analysis of objective information and is explained by expected utility theory. However, less researcher consider information effect on travelers' psychological feelings. In this paper, objective information affecting route choice is firstly analyzed and prospect theory is applied to do psychological mapping of the objective information of route choice contributing factors so as to forming subjective information. Then, the relationship model between the subjective information and evaluation levels is proposed. A BP fuzzy neural network method is designed to simulate relationship model and solve it. Route choice possibility evaluation value is obtained by fuzzy aggregation and defuzzification processing of evaluation levels, which reflects probability of traveler's route choice. Finally, a real network is used to test the validation of this model.

Hyperspectral images are captured from hundreds of narrow and contiguous bands from the visible to infrared regions of electromagnetic spectrum. Due to the presence of large number of bands, classification of hyperspectral images becomes computation intensive. In this paper, attempt has been made to develop a supervised feature selection technique guided by evolutionary algorithms. Cloud model based self-adaptive differential evolution (CMSaDE) is used for feature subset generation. Generated subsets are evaluated using a wrapper model where fuzzy k-nearest neighbor classifier is taken into consideration. Our proposed method also uses a feature ranking technique, ReliefF algorithm, for removing duplicate features. To demonstrate the effectiveness of the proposed method, investigation is carried out on KSC data set and the results are compared with four other evolutionary based state-of-the-art feature selection techniques. The proposed method shows promising results compared to others in terms of overall classification accuracy and Kappa coefficient.

This paper presents a robust system for online handwritten Chinese/Japanese character recognition. Online handwritten character recognition, recognizing characters from their trajectories, will be used more widely, with the development and proliferation of pen-based or touch-based input devices such as tablet terminals, smart phones, electronic whiteboards and digital pens (e.g., Anoto pen). This paper focuses on the online handwritten Chinese/Japanese character recognition, and describes its recent technology trends, problems and methods to solve them.

The technology of multi-focus imaging is to fuse multiple images, which have the same scene but with difference focal distance, into one picture with several focusing simultaneously. General multi-focus image fusion technology can only process gray image mostly. In this paper, we process color images and use color saturation based on chrominance divided by luminance to do multi-focusing. The focus is calculated by deploying a star-light shaped mask. To deal with object displacement or deformation phenomena, image registration is necessary before fusion. The obtained mean correction rate and PSNR are 80% and 42dB for tested color images, respectively. For test gray images, the obtained mean correction rate and PSNR are 83% and 40dB, respectively. The experimental results proved that the proposed color saturation based star-light focusing detection can be applied to image fusion reliably.

Fingerprint classification is very important for improving the efficiency of fingerprint identification. In some sense, the process of fingerprint classification is also a rough process of fingerprint matching. This article discusses neural network and pattern identification technique in a combined manner, and proposes a kind of fingerprint classification algorithm based on genetic neural network. Judging from the simulation result, the algorithm could make basically accurate classification on fingerprint, and could reduce the time of fingerprint matching and search as well as calculation complexity. It increased the robustness of the identification system, and was conducive to enhancing the speed and anti-interference ability of fingerprint identification.

Finding human cognitive state is very important and useful for medical cares in our daily life. Eye state classification is a kind of common time-series problem for detecting human cognitive state. One of approaches of EEG eye state classification is based on time series data. We propose a new neural fuzzy structure that is possible to use an eye state classification. The classification accuracy can be achieved by using the proposed approach in terms of the number of neurons in the hidden layer, which also leads types of membership function in fuzzy rules. The data of EEG signals for monitoring eye state saved by Machine Learning Repository, University of California, Irvine (UCI) is used for eye state benchmarking. We did experiments with 3 different networks for the architecture by changing the number of neurons in the hidden layer and random seed for weights. We found that tuning parameters asymmetrically gave us the best results through test cases. According to the test results, we have the best result with 4 neurons by managing parameters of standard deviations asymmetrically, with which showed a 4.0% average error rate with the test data.

With the development of computer graphics technology, machine vision and virtual reality technology in recent years, 3D reconstruction method through the sequence of images of outdoor scenes has become a key research direction in computer vision and graphics. During the acquisition process of image, due to the measurement equipment and environment, single shot sequence of photos may not be able to extract enough surface information and lead to unable to complete the reconstruction of 3D objects. To solve this problem, fusion method of point clouds from multi-group images is adopted in the thesis. Firstly the color histogram matching is used to complete the supplement image sequence. Next the point cloud from the supplement image sequence is solely calculated. Then the transform parameters in overlap area of different point clouds are computed by using the improved iterative closest point algorithm. Finally the registration and fusion among the point cloud data of different photos is conducted. The experiments show that this method can effectively supplement point cloud data for reconstruction.

Compared to the dimension of face image samples, the number and face image is relatively small. The face recognition problem is essentially a small sample learning problem. Aiming at the small sample problem, in this paper, we propose the method of self-training for margin neighbor. Using the margin to represent decision confidence, and using the spatial adjacent to represent gradual change of face manifold, through self-training iteration, the sample distance of the same classifications is as compact as possible, the sample distance of the different classifications maintain a certain large distance. In the neighborhood, constantly mark the unlabeled samples of high credibility. Experiments show that, compared to other methods, self-training for large margin neighbor has relatively better recognition in small face samples.

With the rapid popularity of the 3D-TV, it is becoming more and more significant to protect the copyright in 3D image. To cater this need, we proposed a novel digital watermarking of 3D image scheme based on visual cryptography. Anaglyph 3D image is separated into left eye view image, right eye view image and the depth view image, while the watermark is pretreated based on the visual cryptography. Watermarking content protection is done with the help of Discrete Wavelet Transform (DWT) on anaglyph 3D images.

Manifold Learning algorithms are significant and have been popularly applied in artificial intelligent, image processing and computer vision applications. In this paper, we propose a novel manifold learning algorithm for image classification. It improves the locally linear embedding (LLE) algorithm, namely imp-LLE. The algorithm integrates the identification to preferably improve the optimation aiming at keeps the intrinsic topology composition of the initiate data. We apply the proposed imp-LLE algorithm into image classification visualization, which attains a better classification visualization result compared with existing manifold learning algorithms.

Left-linear grammar and right-linear grammar are known collectively as regular grammar, which defines a description mechanism for lexical analyzing. Meanwhile Finite Automata (FA) provides a recognition mechanism for tokens, which can be constructed from regular grammar. This paper proposed a conversion method between left-linear grammar and right-linear grammar via finite automata. Firstly, one of them is converted to finite automata, and then rewrites the grammar based on the constructed finite automata.

Through the study of MEMS packaging technology, the quartz tuning fork crystal oscillator is directly capsulated in MEMS vacuum packaged shell and the phase and amplitude comparison circuits are used to measure and monitor the degree of vacuum in the internal cavity of the MEMS package. The resistance welding method for vacuum packaging and the vacuum measurement technologies have been developed to improve the performance of the micromachined quartz tuning fork gyroscopes. The threshold is enhanced about 10 times.

The microprocessor differential protection is the main protection of single-phase traction transformer in the traction power supply system. The conventional algorithm based on differential protection is often unable to identify the magnetizing inrush current correctly, and causes the malfunction. The artificial neural network with the adaptive ability, based on the experimental data and the field data, can identify the magnetizing inrush current and the short-circuit current. This method is accurate, and has no misjudgment phenomenon.

Through the chip YT5188 and Trident 2SA695 setting up a circuit, this article mainly use a DC motor as a control object. A mechanical position sensor is driven by a bridge type circuit (potentiometer), composed of a digital feedback control system, forming the position closed-loop, so as to form a position controller, adjusting the potentiometer resistance to the effect of servo control. Through the method of PWM pulse width modulation, the Angle of the DC motor output can be controlled with a more convenient location.

In order to overcome the weakness of BP algorithm, a new type of lithium battery SOC estimation method which is based on particle swarm optimization and the BP neural network is proposed. This method uses network training error as particle swarm optimization's fitness value and then iteration to find the optimal individual as the network initialization thresholds and weights. The average error of the SOC estimation is 1% when compared with the real SOC obtained from a discharge test. This show higher accuracy compared with BP neural network.

This paper research a method that can confirm the software evolution based on Latent Dirichlet Allocation (LDA). LDA is a method that can analyze the interdependency among words, topics and documents, and the interdependency can be expressed as probability. In this paper, adoption of LDA to modeling software evolution, take the package in source code as a document, regard names of function (method), variable names and comments as words, and figure out the probability between the three. Take results compare with update reports, can confirm the software of new version consistent with update reports.

In the field of application of parallel storage, the increased demand and system function expansion have put forward a higher requirement on the storage performance of reading and writing. On the basis of VxWorks, this paper makes a deep research of dos File System (dosFs) and raw File System (rawFs). And we found that the reading and writing speed of Solid State Disk (SSD) under rawFs is faster than it under dosFs. What's more, we proposed an algorithm named Dynamic self-Adaptation Threshold Algorithm (DATA) which takes advantage of the free time to write back some data into data buffer. The algorithm can reduce the waiting time so that it can increase the performance of SSD. In order to further increase the reading and writing speed, we applied this algorithm to rawFs. Through the experiment result we reach the conclusion that the performance of SSD under rawFs is better than it under dosFs. And the performance of SSD under rawFs which has been used DATA is better than it under rawFs which hasn't been used.

In order to solve information island problems and promote information sharing among multiple information systems in Chinese railway, it is necessary to improve railway master data management (MDM) to ensure the identification uniqueness, characteristics consistency of master data. At present, it is lack of planning and standard, separate collection and maintenance for MDM in railway. In this paper, the railway master data planning is given and the life cycle management of railway master data is analyzed. On this basis, the system architecture of railway MDM platform is presented and the MDM platform is designed. At last the benefits of master data management for Chinese railway are analyzed.

The fuzzy classification plays an important role to predict defect of software modules. In this paper, the fuzzy measure (FM) is used to improve the predict accuracy and capability by acquiring all possible interactions among metrics and apply Choquet integral (CI) for classifying in n dimensional space and automatic searching the least misclassification rate based on distance. To implement the model, we also need to determine the unknown parameters, and which is implemented using genetic algorithm (GA) on the training data. The proposed model is tested on the four NASA software projects. The results indicate that the predict performance of proposed model is better than other predict models.

In order to improve the quality of the Java source code for meeting the design requirements of the “high cohesion, low coupling”. This paper proposes the extended Jaccard coefficient for measuring the similarity between the class not only considering the inheritance, association etc, but also the method caller and called relationship. Also we implement a prototype system to aid software maintains by clustering several closely related classes into the same package. The tests show that the structure of java source code is the more in line with the characteristics of high cohesion and low coupling after adjustment.

Firstly, the theory and techniques of the CET4 Diagnostic Practice System were analyzed. Secondly, the architecture and development mode of the system has been determined and the main functional modules has been designed detailedly. Last, the implementation of system architecture has been elaborated on the basis of the development environment and tools. This system is a kind of deep practice and innovation on CET-4, aimed at learners can receive personalized learning guidance and targeted diagnostic evaluation according to their characteristics, thus improving the standard of English learners and learning ability, so it has a wide range of prospect and practical value.

The user's identity management which becomes increasingly burdensome can be finished and the safety and reliability of the user's identity identification of the software system can be improved in virtue of RBAC model. RBAC module shall also be properly changed and modified based on the specific software application system to better adapt to and meet the business demand. Improved RBAC module is proposed in this paper based on the demand of the project construction enterprise through the study of the actual development process of the user's identity authority management in the safety quality system of project construction and the RBAC module-based workflow management system is also analyzed. In the operation process of the actual project, improved RBAC-based workflow platform has been successfully researched and developed and put into the project construction line. It plays a vital role in connecting several business systems and realizing the safety management of the user's identity in the management safety quality management of the project.

The following sections is included:

• Author Index